For as long as the IT industry has been in existence, one of its major purposes has been to automate tasks routinely carried out by human workers. There is no doubt that properly constructed applications can carry out many tasks on a 24 hour basis without error and at very little cost. However, it may be possible that we are taking this assumption one step too far.
Nobody is questioning the automation of simple repetitive tasks. However, as our applications become more complex, we are attempting more and more to simulate the activity of the human brain - building intelligence into our software. This is very much the case with enterprise management solutions that have focussed very closely on self-healing and pattern matching of events within their solutions. However, it is possible that the complexity and lack of accuracy within these components do not create the value that is expected and, consequently, the all-important return on investment is much more difficult to achieve.
First, Computer Associates with its Neugent technology and, more recently, IBM with its Autonomic initiatives and HP with Adaptive technology are attempting to carry out a much deeper analysis of events within the managed environment. This is an appropriate evolution of the technology and is exciting to watch as it unfolds. However, is it really useful for the customer?
Often, the best solution is to present the information to a human operator in a way that utilises the recognition skills of the human brain. An excellent example of this is the BEST/1 performance management solution, originally developed by BGS and now a part of the BMC product portfolio. In complex scenarios, this technology simply presents coloured pixels that indicate the status of individual components. Problem areas show up as patches of colour and are instantly recognisable to the human eye. Analysing the raw data at the same speed would require massive computing resources.
The issue is whether the tasks being automated are being carried out successfully and also whether a human operator might be able to arrive at a better conclusion more quickly. Intelligent software is expensive and likely to be inaccurate. This contradicts our original premise for automation. If we add to the equation the falling cost of human skills, resulting from weaker job markets and increased use of resources in India and other low cost areas or the World, the cost justification for this type of solution is more difficult to see.
